Hi, My experiment consists of fast-rate events (min onset to onset interval 3.1 s) (6 EVs, all with double-gamma HRF and temporal filtering&derivative). As the stimuli are 100 ms in duration, it seems logical to enter the timings using the 3 column format as follows: time 0.1 1 However, in this case, Feat complains about a rank deficent design. If I now change the duration for all events from 0.1 to 0.5, time 0.5 1, there will be no error messages (and the results of the analysis seem ok). How could the design matrix be less rank deficient when I change the event duration from 0.1 to 0.5? Is the duration of 0.1s somehow too short for Feat to put up the model? Or could it be, for some reason, that the events in my experiment are too tightly packed? -- Teemu Rinne Dept Psych/Univ Helsinki